Conjunct COST B27 and SAN Scientific Meeting, Swansea, UK, 16-18 September 2006.

Slides:



Advertisements
Similar presentations
Clinical Applications of Spectral Analysis Winni Hofman, PhD University of Amsterdam Medcare Amsterdam.
Advertisements

Brain Wave Based Authentication
THE EEGEEG James Peerless April Objectives Physics and Clinical Measurement Anaesthesia for neurosurgery, neuroradiology and neurocritical care.
Analysis of variance (ANOVA)-the General Linear Model (GLM)
Poster Design & Printing by Genigraphics ® Leonard J. Trejo, Ph. D. Roman Rosipal, Ph. D Pacific Development and Technology, LLC Paul L.
Model Adequacy Checking in the ANOVA Text reference, Section 3-4, pg
Principals of Digital Signal Recording. How do we represent a continuously variable signal digitally? Sampling – Sampling rate – number of measurements.
Induced Brain Waves by Binaural Beats: A Study on Numerosity.
LFPs 1: Spectral analysis Kenneth D. Harris 11/2/15.
Pre-processing for EEG and MEG
Environmental Data Analysis with MatLab Lecture 23: Hypothesis Testing continued; F-Tests.
Analysis of frequency counts with Chi square
Abstract  Obstructive Sleep Apnea Syndrome (OSAS) is a very common sleep disorder with potential severe implications in essential aspects and the patient's.
Environmental Data Analysis with MatLab Lecture 24: Confidence Limits of Spectra; Bootstraps.
General Linear Model & Classical Inference
Wavelet transformation Emrah Duzel Institute of Cognitive Neuroscience UCL.
ElectroScience Lab IGARSS 2011 Vancouver Jul 26th, 2011 Chun-Sik Chae and Joel T. Johnson ElectroScience Laboratory Department of Electrical and Computer.
WASSIM NASREDDINE MD AUBMC EEG: New Applications in Psychiatry?
Comodulation and Coherence in Normal and Clinical Populations
WAVELET TRANSFORM.
Statistical Decision Making. Almost all problems in statistics can be formulated as a problem of making a decision. That is given some data observed from.
General Linear Model & Classical Inference London, SPM-M/EEG course May 2014 C. Phillips, Cyclotron Research Centre, ULg, Belgium
Measures of Dispersion & The Standard Normal Distribution 2/5/07.
1 Methods for detection of hidden changes in the EEG H. Hinrikus*, M.Bachmann*, J.Kalda**, M.Säkki**, J.Lass*, R.Tomson* *Biomedical Engineering Center.
SPM short course – Oct Linear Models and Contrasts Jean-Baptiste Poline Neurospin, I2BM, CEA Saclay, France.
Functional Brain Signal Processing: EEG & fMRI Lesson 4
Signal and System I The unit step response of an LTI system.
 Difficult to measure consciousness.  A state of consciousness is referred to as a hypothetical construct.  I.e., a concept used to describe something.
Aaron Raymond See Department of Electrical Engineering Southern Taiwan University 11/17/20151.
PSYC 6130 One-Way Independent ANOVA. PSYC 6130, PROF. J. ELDER 2 Generalizing t-Tests t-Tests allow us to test hypotheses about differences between two.
Intrinsic Short Term Variability in W3-OH and W49N Hydroxyl Masers W.M. Goss National Radio Astronomy Observatory Socorro, New Mexico, USA A.A. Deshpande,
Spectral Analysis of Resting State Electroencephalogram (EEG) in Subjects With and Without Family Histories of Alcoholism Spectral Analysis of Resting.
GG313 Lecture 24 11/17/05 Power Spectrum, Phase Spectrum, and Aliasing.
References 1.Schnider TW et al, Anesthesiology 1998;88: Minto CF et al, Anesthesiology 2003;99: Smith WD et al, Stat Med 1996;15:
Direct electrical cortical stimulation to reconstruct epileptiform afterdischarge networks Lorenzo Caciagli 1, Roman Rodionov 1, Catherine Scott 2, Tim.
Principal components analysis (PCA) as a tool for identifying EEG frequency bands: I. Methodological considerations and preliminary findings Jürgen Kayser,
What to Measure Sampling and generalizability  Population vs. sample  Sampling techniques – procedures for deciding which examples of the population.
Jump to first page Inferring Sample Findings to the Population and Testing for Differences.
Université d’Ottawa / University of Ottawa 2001 Bio 8100s Applied Multivariate Biostatistics L11.1 Lecture 11: Canonical correlation analysis (CANCOR)
An ERP investigation of response inhibition in adults with DCD Elisabeth Hill Duncan Brown José van Velzen.
NURS 306, Nursing Research Lisa Broughton, MSN, RN, CCRN RESEARCH STATISTICS.
Computer Architecture and Networks Lab. 컴퓨터 구조 및 네트워크 연구실 EEG Oscillations and Wavelet Analysis 이 윤 섭이 윤 섭.
Methods for Dummies M/EEG Analysis: Contrasts, Inferences and Source Localisation Diana Omigie Stjepana Kovac.
Stats Methods at IC Lecture 3: Regression.
Fourier series With coefficients:.
k is the frequency index
Neurofeedback of beta frequencies:
General Linear Model & Classical Inference
Aaron Raymond See Department of Electrical Engineering
Advanced applications of the GLM: Cross-frequency coupling
Advanced applications of the GLM: Cross-frequency coupling
Measures of Complexity
Ketamine increases the frequency of electroencephalographic bicoherence peak on the α spindle area induced with propofol  K. Hayashi, N. Tsuda, T. Sawa,
Electroencephalography and delirium in the postoperative period
Perceptual Echoes at 10 Hz in the Human Brain
Human neural correlates of sevoflurane-induced unconsciousness
k is the frequency index
ERRORS, CONFOUNDING, and INTERACTION
Ch10 Analysis of Variance.
Cycle 10: Brain-state dependence
Dynamic Causal Modelling for M/EEG
8.5 Modulation of Signals basic idea and goals
Machine Learning for Visual Scene Classification with EEG Data
Electroencephalography and delirium in the postoperative period
Comparison of Bispectral Index and Entropy values with electroencephalogram during surgical anaesthesia with sevoflurane†  A.J. Aho, K. Kamata, V. Jäntti,
Volume 75, Issue 5, Pages (September 2012)
Identification of phasic activity bursts in the dentate gyrus.
Advanced applications of the GLM: Cross-frequency coupling
Volume 66, Issue 1, Pages (April 2010)
Presentation transcript:

Conjunct COST B27 and SAN Scientific Meeting, Swansea, UK, 16-18 September 2006

Bispectral analysis of the EEG: what does it add to the state versus non-state debate in hypnosis? Adrian Burgess, University of Swansea Helen Crawford, Virginia Polytechnic Institute & State University

Plan of Talk What is bispectral analysis? State –vs- Non-state theories of Hypnosis Why is bispectral analysis relevant to the State –vs- Non-state debate? The EEG bispectrum in hypnosis and waking for high and low susceptible participants

What is bispectral analysis? Bispectral analysis is a Fourier based method for examining the coupling between frequencies in different ranges The bispectrum is defined as: Where X(.)=Fourier Transform of the time series x(t) and * indicates the complex conjugate Bicoherence is the normalised bispectrum:

What is bispectral analysis? Within Channels Between Channels 1st order Mean 2nd order Fourier Spectrum Coherence 3rd order Bicoherence Cross-bicoherence

State -vs- Non-state theories of Hypnosis State theorists believe that hypnosis is an altered state of consciousness, Non-state theorists believe that hypnotic effects are the product of more-mundane psychological processes such as expectancy & role-play

Neurophysiological evidence in favour of the State-Theory More than 20 years of EEG/ERP research has shown that the hypnotic state is associated neurophysioloigcal changes in Alpha Theta Gamma ERP (e.g. MMN, Somatosensory ERP) etc…. However, the differences are quantitative not qualitative cf other states of consciousness within the normal range

Why is Bispectral Analysis relevant to the State -vs- Non-state debate? Bispectral Analysis has been shown to be a useful measure of level of consciousness ~1000 research papers on Bispectral Analysis and anaesthesia The Bispectral Index (BIS®) is a patented technology produced by Aspect Medical Systems that uses the bicoherence in the EEG the ratio of EEG power in the delta (1–4 Hz) and beta (13–30 Hz) frequency ranges the proportion of the EEG that is isoelectric (i.e. electrical silence) to produce an index of depth of ‘hypnosis’

Hypotheses Participants with high susceptibility to hypnosis will show a significant change in the bispectrum of their EEG between the waking and hypnotic states Participants with low susceptibility to hypnosis will NOT show a significant change in the bispectrum of their EEG between the waking and hypnotic states

Method - Participants Healthy, young, right-handed volunteers Pre-selected using the Stanford Hypnotic Susceptibility Scale (SHSSC) 12 high susceptible (SHSS-C ≥9) Age range 20-24 10 women, 2 men 12 low susceptible (SHSS-C ≤4) 9 women, 3 men

Method- EEG Recorded EEG from young, healthy volunteers 32-channel Neuroscan Synamps 28 EEG Channels Sampling rate 500Hz Bandpass 0.1-150Hz

Method- Procedure Waking SHSS-C Hypnotised Pre-induction Eyes Closed Stanford Hypnotic Susceptibility Scale Pre-induction Eyes Closed Waking Pre-induction Memory test Hypnotic Induction SHSS-C Post-induction Eyes Closed Hypnotised Post-induction Memory test

Calculation of the bispectrum Bispectrum was calculated on the Eyes Closed Condition in Waking (pre-induction) Hypnosis (post-induction) Calculated using the MATLAB toolbox ‘Higher Order Spectral Analysis’ Averaged Bispectrum from the mean of ~4 minutes of EEG divided into epochs of 1.024s Range 0-100Hz with a resolution of ~1Hz.

Example of an EEG bispectrum Alpha Peak (10Hz,10Hz) Alpha-Delta Coupling (8Hz,2Hz) Delta Peak (2Hz,2Hz)

Topography of the bispctrum

Bispectrum by Group and Condition

Partial Least Squares Regression A combination & extension of: Multiple Regression PCA Designed to identify simultaneously Whether the experimental design has an effect Where in the data the effect is seen Used rotated PLS with Hypnosis -vs- Waking For High and Low susceptible groups 1000 randomizations 1000 bootstrap samples Output Latent variables showing contrasts i.e. is there an effect? Saliences showing location of differences i.e where is the effect From Lobaugh et al., 2000

LV 1; 95.3% cross-block variance, p<0.01 1st Latent variable – PLS of Bispectrum LV 1; 95.3% cross-block variance, p<0.01 HIGHS LOWS

Topography of reliable differences between Waking & Hypnosis Bispectrum higher in the Waking condition Midline frontal Temporo-occipital

Reliable differences between Waking & Hypnosis – across all electrodes Bispectrum higher in the Waking condition

Summary PLS analysis showed significant differences in the bispectrum between waking and hypnosis for the High Susceptible group Bispectrum was higher in the waking condition esp at high frequencies Midline frontal Temporo-occipital sites What about bicoherence?

LV 1; 64.6% cross-block variance, p<0.26 1st Latent variable – PLS of Bicoherence LV 1; 64.6% cross-block variance, p<0.26 HIGHS LOWS

Why the discrepancy? The only difference between the bispectrum and bicoherence is the normalisation Normalisation is by the power in the signal at the relevant frequencies Therefore, the differences between Waking and Hypnosis must be in the Fourier Spectrum However, with very low power levels, esp at high frequencies, normalisation can give erroneous estimates of bicoherence

LV 1; 87.3% cross-block variance, p<0.045 1st Latent variable – PLS of Fourier Spectrum LV 1; 87.3% cross-block variance, p<0.045 HIGHS LOWS

Reliable differences in the Fourier Spectrum by frequency Band Left-right difference Ant-Post difference Global difference Midline Parietal difference Ant-Post difference Ant-Post difference RED: Waking>Hypnosis BLUE: Hypnosis>Waking

Reliable differences in the Fourier Spectrum by frequency

Summary Frequency Band Waking > Hypnosis Hypnosis > Waking Delta Right side Left Side Theta Frontal Midline Occipital Alpha Global - Beta Midline Parietal Gamma

Final Summary High susceptibles show significantly greater Bispectral values in the waking condition than in hypnosis, esp High frequencies Midline frontal Temporo-occipital sites There are no differences in Bicoherence The differences in the Bispectrum are due to differences in the power spectra of the EEG Calculation of the Bispectrum is problematic

Conclusion Q. Does bispectral analysis add anything to the state versus non-state debate in hypnosis apart from complexity? Probably not But, with improved estimation of bicoherence it might But, PLS analysis of the FFT was helpful in elucidating the EEG power differences seen between the waking and Hypnotic states seen in High Susceptibles Ho hum

Thank you